Bug 127479 – [Win] Whitespace differences in test results for fast/{css, dom, forms, multicolor, ruinin}

• Status: NEW
• Component: Tools / Tests

[Daniel Bates]

As of the time of writing, the output of the following layout tests differ in whitespace on the Windows Debug and Release bots when compared to other ports, including the Apple Mac port:

fast/css/first-letter-block-form-controls-crash.html
fast/dom/adopt-node-crash-2-expected.html
fast/forms/access-key-for-all-elements-expected.html
fast/forms/form-associated-element-crash3-expected.html
fast/multicol/renderer-positioned-assert-crash-expected.html
fast/runin/nonblock-runin-expected.html

Windows Debug Tests:
<http://build.webkit.org/results/Apple%20Win%207%20Debug%20(Tests)/r162605%20(57440)/results.html>

Windows Release Tests:
<http://build.webkit.org/results/Apple%20Win%207%20Release%20(Tests)/r162605%20(41965)/results.html>

[Daniel Bates]

Disregarding the whitespace differences, these tests are all passing. As discussed with in-person with Brent Fulgham, yesterday (01/22), I'll land Windows-specific results for these tests for now as opposed to skip them so that we can identify a progression. We should further investigate such whitespace differences.

[Daniel Bates]

Created attachment 221978 [details]
Whitespace difference in results

For convenience, I've attached a diff between the expected results for the aforementioned tests and the Windows-specific results.

[Daniel Bates]

Committed Windows-specific expected results in r162607, <http://trac.webkit.org/changeset/162607>.

[Daniel Bates]

The test results for fast/css/relative-position-replaced-in-table-display-crash.html also differs due to whitespace differences.

Committed Windows-specific expected result in <http://trac.webkit.org/changeset/162620>.